Generating Efficient DNN-Ensembles with Evolutionary Computation

TL;DR

This paper introduces EARN, an evolutionary method for creating efficient DNN ensembles that optimize accuracy, inference time, and model size, outperforming existing approaches on multiple datasets.

Contribution

We propose EARN, a novel evolutionary approach that effectively optimizes DNN ensembles for multiple objectives, reducing evaluation costs and improving model efficiency.

Findings

Achieved up to 7.60x speedup in inference time.

Reduced model parameters by up to 10x.

Improved accuracy by up to 6.01% over the best individual DNN.

Abstract

In this work, we leverage ensemble learning as a tool for the creation of faster, smaller, and more accurate deep learning models. We demonstrate that we can jointly optimize for accuracy, inference time, and the number of parameters by combining DNN classifiers. To achieve this, we combine multiple ensemble strategies: bagging, boosting, and an ordered chain of classifiers. To reduce the number of DNN ensemble evaluations during the search, we propose EARN, an evolutionary approach that optimizes the ensemble according to three objectives regarding the constraints specified by the user. We run EARN on 10 image classification datasets with an initial pool of 32 state-of-the-art DCNN on both CPU and GPU platforms, and we generate models with speedups up to $7.60\times$, reductions of parameters by $10\times$, or increases in accuracy up to $6.01\%$ regarding the best DNN in the pool. In addition, our method generates models that are $5.6\times$ faster than the state-of-the-art methods for automatic model generation.